Bongartz, Tim

Objective To test the ability of machine learning (ML) approaches with clinical and genomic biomarkers to predict methotrexate treatment response in patients with early rheumatoid arthritis (RA). Methods Demographic, clinical, and genomic data from 643 patients of European ancestry with early RA (mean age 54 years; 70% female) subdivided into a training (n = 336) and validation cohort (n = 307) were used. The genomic data comprised 160 single‐nucleotide polymorphisms (SNPs) previously associated with RA or methotrexate metabolism. Response to methotrexate monotherapy was defined as good or moderate by the European Alliance of Associations for Rheumatology (EULAR) response criteria at the 3‐month follow‐up. Supervised ML methods were trained with 5 repeats and 10‐fold cross‐validation using the training cohort. Prediction performance was validated in the independent validation cohort. Results Supervised ML methods combining age, sex, smoking, rheumatoid factor, baseline Disease Activity Score in 28 joints (DAS28) scores and 160 SNPs predicted EULAR response at 3 months with the area under the receiver operating curve of 0.84 ( P = 0.05) in the training cohort and achieved a prediction accuracy of 76% ( P = 0.05) in the validation cohort (sensitivity 72%, specificity 77%). Intergenic SNPs rs12446816, rs13385025, rs113798271, and ATIC (rs2372536) had variable importance above 60.0 and along with baseline DAS28 scores were among the top predictors of methotrexate response. Conclusion Pharmacogenomic biomarkers combined with baseline DAS28 scores can be useful in predicting response to methotrexate in patients with early RA. Applying ML to predict treatment response holds promise for guiding effective RA treatment choices, including timely escalation of RA therapies.

Abstract We previously reported, on the basis of a genome-wide association study for aromatase inhibitor-induced musculoskeletal symptoms, that single-nucleotide polymorphisms (SNPs) near the T-cell leukemia/lymphoma 1A (TCL1A) gene were associated with aromatase inhibitor-induced musculoskeletal pain and with estradiol (E2)-induced TCL1A expression. Furthermore, variation in TCL1A expression influenced the downstream expression of proinflammatory cytokines and cytokine receptors. Specifically, the top hit genome-wide association study SNP, rs11849538, created a functional estrogen response element (ERE) that displayed estrogen receptor (ER) binding and increased E2 induction of TCL1A expression only for the variant SNP genotype. In the present study, we pursued mechanisms underlying the E2-SNP-dependent regulation of TCL1A expression and, in parallel, our subsequent observations that SNPs at a distance from EREs can regulate ERα binding and that ER antagonists can reverse phenotypes associated with those SNPs. Specifically, we performed a series of functional genomic studies using a large panel of lymphoblastoid cell lines with dense genomic data that demonstrated that TCL1A SNPs at a distance from EREs can modulate ERα binding and expression of TCL1A as well as the expression of downstream immune mediators. Furthermore, 4-hydroxytamoxifen or fulvestrant could reverse these SNP-genotype effects. Similar results were found for SNPs in the IL17A cytokine and CCR6 chemokine receptor genes. These observations greatly expand our previous results and support the existence of a novel molecular mechanism that contributes to the complex interplay between estrogens and immune systems. They also raise the possibility of the pharmacological manipulation of the expression of proinflammatory cytokines and chemokines in a SNP genotype-dependent fashion.

Abstract Background: The use of anti-estrogen (AE) therapies (selective estrogen receptor modulators [SERMs], aromatase inhibitors [AIs]) in breast cancer (BC) patients was recently associated with risk of developing rheumatoid arthritis (RA) (J Rheum 2015;42:55-9). We attempted to replicate and extend these results using electronic health record (EHR) enhanced cancer registry data that assessed specific AE therapies and also accounted for potential confounding by age, smoking status, and chemotherapy use. Methods: Using cancer registry and EHR data, we identified a cohort of BC patients who were newly diagnosed and treated at Mayo Clinic Rochester from 1998-2011 and had no previous diagnosis of RA. Smoking status at diagnosis, BC treatments and RA at diagnosis and during follow-up were identified using validated EHR-based algorithms. Hazard ratios (HRs) and 95% confidence intervals (CI) from a multivariate Cox model were used to estimate the association of AE use with risk of RA, controlling for age, smoking status at BC diagnosis, and BC chemotherapy use. Results: The analytic cohort of 9,244 newly diagnosed BC patients treated and followed at Mayo Clinic had a median age at diagnosis of 59 years (range 18-97); 32% were smokers; and 29% were treated with chemotherapy. During a median follow-up of 49 months (range 1-191), 19 patients developed RA. Compared to BC patients not receiving any AE therapy, those receiving AI monotherapy, but not SERM monotherapy or both SERMs and AIs, were at significantly increased risk of developing RA (Table). AE Therapy# of RA Cases / # patientsHR* (95% CI)P-valueNo AE therapy6/4,0711.00 (reference)SERMs only5/2,5371.20 (0.36 - 3.95)0.77AIs only5/7126.37 (1.77 - 22.92)0.005SERMs then AIs3/1,8750.90AI then SERMs0/490.91 (0.21 - 3.86)*Adjusted for age, smoking status and use of chemotherapy Conclusions: Our findings indicate that the use of AIs for BC therapy, but not SERMs or SERMs followed by AIs, is associated with increased risk for development of RA, when controlling for smoking status and use of chemotherapy. While needing replication, these findings suggest a specific role for AIs in RA development and suggest pathways that could be targeted to prevent this potential treatment complication. Citation Format: Matthew K. Breitenstein, Ming-Fen Ho, Richard M. Weinshilboum, James R. Cerhan, Jyotishman Pathak, Tim Bongartz, Liewei Wang, James N. Ingle. Association of anti-estrogen therapy in breast cancer patients and subsequent risk of rheumatoid arthritis: An electronic health record study. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-191. doi:10.1158/1538-7445.AM2015-LB-191

ObjectivePulmonary disease represents an important extraarticular manifestation of rheumatoid arthritis (RA). While the association of RA and interstitial lung disease is widely acknowledged, obstructive lung disease (OLD) in RA is less well understood. We therefore aimed to assess the incidence, risk factors, and mortality of OLD in patients with RA.MethodsWe examined a population‐based incident cohort of patients with RA and a comparison cohort of individuals without RA. OLD was defined using a strict composite criterion. Cox proportional hazards models were used to compare OLD incidence between the RA and comparator cohorts to investigate risk factors and to explore the impact of OLD on patient survival.ResultsA total of 594 patients with RA and 596 subjects without RA were followed for a mean of 16.3 and 19.4 years, respectively. The lifetime risk of developing OLD was 9.6% for RA patients and 6.2% for subjects without RA (hazard ratio [HR] 1.54, 95% confidence interval [95% CI] 1.01–2.34). The risk of developing OLD was higher among male patients, among current or former smokers, and for individuals with more severe RA. Survival of RA patients diagnosed with OLD was worse compared to those without OLD (HR 2.09, 95% CI 1.47–2.97).ConclusionPatients with RA are at higher risk of developing OLD, which is significantly associated with premature mortality. Effective diagnostic and therapeutic strategies to detect and manage OLD in patients with RA may help to improve survival in these patients.